文章是基于模型平均的方法对我国重点省会城市月度房价数据的空间自回归模型的拓展研究. 通过一定的宏观经济解释变量和房价数据, 构建区域房价的空间自回归模型, 并在基于MMA准则的模型平均框架下, 将不同的候选模型组合进行房价预测. 对比经典空间自回归模型的预测, 基于模型平均的MMA, SAIC和SBIC的预测有更高的精确度和更好的稳定性.

This article is based on the model averaging method, and extends the spatial autoregressive model of the monthly housing prices data of cities in China. A spatial autoregressive model was established by certain macroeconomic explanatory variables and housing prices data. Different candidate models were combined to predict housing prices under the model averaging of MMA criterion. Compared with the prediction of classical spatial autoregressive model, the model averaging predictions of MMA, SAIC and SBIC are more accurate and more stable.